|
On Thu, Jul 06, 2000 at 04:56:14PM -0500, David D.W. Downey wrote:
> Evening folks,
> I've got a few problems here that I just can not figure out.
>
> My set up is as follows. I have UltraMonkey (ultramonkey.sourceforge.net)
> installed. I have the IPs rolling over correctly. (the IPs for the fkae
> server and the NAT device) I'm using a serial null modem connection to
> get that running correctly. The problem that I have is on the load
> balancing.
>
> I have any requests coming in on the fake IP on the internet side sending
> every single request to the same machine. I have 3 total machines listed
> and only one being hit.
> ldirectord.cf
> ============
> timeout=3
> checkinterval=1
> fallback=127.0.0.1:80
> virtual=216.200.192.111:80
> real=192.168.1.12:80 masq
> real=192.168.1.10:80 masq
> real=192.168.1.11:80 masq
> service=http
> request="index.htm"
> receive="QIXO"
> scheduler=rr
> #persistent=600
> protocol=tcp
>
> (NOTE: Bear in mind that "fallback" is not running HTTP, I just filled it in
> so it would quit bitching.)
> I can not figure out what in the hell the problem is. No matter what I
> do, including rebooting, I can not clear out the entries listed in
> ipchains -M -L. It continuosly shows ws-02 as being the only allowed
> entry.
What is the output of "ipvsadm -L -n". If it sows only one real server then
I suspect that the problem lies in ldirectord monitoring the real servers.
You should check that the output of http://192.168.1.12:80/index.htm,
http://192.168.1.10:80/index.htm and http://192.168.1.11:80/index.htm all
contain the string QIXO. A quick test to verify this would be to runn the
following on the LVS server.
lynx -dump http://192.168.1.10:80/index.htm | fgrep QIXO
lynx -dump http://192.168.1.11:80/index.htm | fgrep QIXO
lynx -dump http://192.168.1.12:80/index.htm | fgrep QIXO
You can see what servers ldirectord is adding into and out of the available
pool (output of ipvsadm -L -n) by examining /var/log/ldirectord.log.
--
Horms
|
